ADD: Case insensitive text search in UTF-16 files

src/fFindDlg.pas

@@ -589,20 +589,6 @@ var
   AEncoding: String;
 begin
   AEncoding:= NormalizeEncoding(cmbEncoding.Text);
-  if (AEncoding = EncodingUCS2LE) or (AEncoding = EncodingUCS2BE) then
-    begin
-      if cbCaseSens.Enabled then
-      begin
-        cbCaseSens.Tag:= Integer(cbCaseSens.Checked);
-        cbCaseSens.Checked:= True;
-        cbCaseSens.Enabled:= False;
-      end;
-    end
-  else if not cbCaseSens.Enabled then
-    begin
-      cbCaseSens.Checked:= Boolean(cbCaseSens.Tag);
-      cbCaseSens.Enabled:= True;
-    end;
   cbTextRegExp.Enabled := (AEncoding = EncodingAnsi);
   if not cbTextRegExp.Enabled then cbTextRegExp.Checked:= False;
 end;
@@ -829,7 +815,20 @@ end;
 
 procedure TfrmFindDlg.cbTextRegExpChange(Sender: TObject);
 begin
-  if cbTextRegExp.Checked then cbCaseSens.Checked:= False;
+  if cbTextRegExp.Checked then
+    begin
+      if cbCaseSens.Enabled then
+      begin
+        cbCaseSens.Tag:= Integer(cbCaseSens.Checked);
+        cbCaseSens.Checked:= False;
+        cbCaseSens.Enabled:= False;
+      end;
+    end
+  else if not cbCaseSens.Enabled then
+    begin
+      cbCaseSens.Checked:= Boolean(cbCaseSens.Tag);
+      cbCaseSens.Enabled:= True;
+    end;
 end;
 
 procedure TfrmFindDlg.cbSelectedFilesChange(Sender: TObject);

src/fviewer.pas

@@ -2025,6 +2025,15 @@ begin
         bTextFound := (PAnsiAddr <> Pointer(-1));
         if bTextFound then FLastSearchPos := PAnsiAddr - ViewerControl.GetDataAdr;
       end
+      // Using special case insensitive UTF-16 search algorithm
+      else if (ViewerControl.Encoding in [veUtf16le, veUtf16be, veUcs2le, veUcs2be]) then
+      begin
+        PAnsiAddr := PosMemW(ViewerControl.GetDataAdr, ViewerControl.FileSize,
+                             FLastSearchPos, sSearchTextA, bSearchBackwards,
+                             ViewerControl.Encoding in [veUtf16le, veUcs2le]);
+        bTextFound := (PAnsiAddr <> Pointer(-1));
+        if bTextFound then FLastSearchPos := PAnsiAddr - ViewerControl.GetDataAdr;
+      end
       // Using very slow search algorithm
       else if (ViewerControl.Encoding in ViewerEncodingMultiByte) or bSearchBackwards then
       begin

src/ufindfiles.pas

@@ -32,7 +32,7 @@ uses
   Classes, SysUtils, DCBasicTypes, uFile;
 
 type
-  TTextSearch = (tsAnsi, tsUtf8, tsOther);
+  TTextSearch = (tsAnsi, tsUtf8, tsUtf16le, tsUtf16be, tsOther);
   TTimeUnit = (tuSecond, tuMinute, tuHour, tuDay, tuWeek, tuMonth, tuYear);
   TFileSizeUnit = (suBytes, suKilo, suMega, suGiga, suTera);
   TPluginOperator = (poEqual, poNotEqual, poMore, poLess, poMoreEqual, poLessEqual,

src/ufindmmap.pas

@@ -42,6 +42,9 @@ function PosMem(pDataAddr: PChar; iDataLength, iStartPos: PtrInt; const sSearchT
 function PosMemU(pDataAddr: PChar; iDataLength, iStartPos: PtrInt;
                  const sSearchText: String; bSearchBackwards: Boolean): Pointer;
 
+function PosMemW(pDataAddr: PChar; iDataLength, iStartPos: PtrInt;
+                 const sSearchText: String; bSearchBackwards, bLittleEndian: Boolean): Pointer;
+
 {en
    Searches a file for a string using memory mapping.
 
@@ -60,10 +63,12 @@ function FindMmap(const sFileName:String; const sFindData:String; bCase:Boolean;
 
 function FindMmapU(const sFileName: String; const sFindData: String): Integer;
 
+function FindMmapW(const sFileName: String; const sFindData: String; bLittleEndian: Boolean): Integer;
+
 implementation
 
 uses
-  DCOSUtils, UnicodeUtils, LazUTF8, StrUtils;
+  SysUtils, DCOSUtils, UnicodeUtils, LazUTF8, StrUtils, DCStrUtils;
 
 function PosMem(pDataAddr: PChar; iDataLength, iStartPos: PtrInt; const sSearchText: String;
                 bCaseSensitive: Boolean; bSearchBackwards: Boolean): Pointer;
@@ -195,8 +200,86 @@ begin
   end;
 end;
 
-function FindMmap(const sFileName, sFindData:String; bCase:Boolean;
-                  Abort: TAbortFunction):Integer;
+function PosMemW(pDataAddr: PChar; iDataLength, iStartPos: PtrInt;
+  const sSearchText: String; bSearchBackwards, bLittleEndian: Boolean): Pointer;
+const
+  BUFFER_SIZE = 4096;
+var
+  iSize: PtrInt;
+  iLength: Integer;
+  iTextPos: Integer;
+  bSwapEndian: Boolean;
+  sTextBuffer: UnicodeString;
+  sLowerCase: UnicodeString;
+begin
+  Result := Pointer(-1);
+  iLength:= Length(sSearchText);
+  bSwapEndian:= {$IFDEF ENDIAN_BIG}bLittleEndian{$ELSE}not bLittleEndian{$ENDIF};
+  if bSearchBackwards then
+  begin
+    iSize:= iStartPos;
+    if iLength > iSize then Exit;
+    sLowerCase:= PUnicodeChar(Pointer(sSearchText + #0));
+    if bSwapEndian then Utf16SwapEndian(sLowerCase);
+    sLowerCase:= UnicodeLowerCase(sLowerCase);
+
+    // While text size > buffer size
+    while iStartPos > BUFFER_SIZE do
+    begin
+      iStartPos:= iStartPos - BUFFER_SIZE;
+      SetString(sTextBuffer, PUnicodeChar(pDataAddr + iStartPos), BUFFER_SIZE div 2);
+      if bSwapEndian then Utf16SwapEndian(sTextBuffer);
+      sTextBuffer:= UnicodeLowerCase(sTextBuffer);
+      iTextPos:= RPos(sLowerCase, sTextBuffer);
+      if iTextPos > 0 then
+        Exit(pDataAddr + iStartPos + iTextPos * 2 - 2)
+      else begin
+        // Shift text buffer
+        iStartPos:= iStartPos + iLength;
+      end;
+    end;
+    // Process remaining buffer
+    if iLength > iStartPos then Exit;
+    SetString(sTextBuffer, PUnicodeChar(pDataAddr), iStartPos div 2);
+    if bSwapEndian then Utf16SwapEndian(sTextBuffer);
+    sTextBuffer:= UnicodeLowerCase(sTextBuffer);
+    iTextPos:= RPos(sLowerCase, sTextBuffer);
+    if iTextPos > 0 then Result:= pDataAddr + iTextPos * 2 - 2
+  end
+  else begin
+    iSize:= iDataLength - iStartPos;
+    if iLength > iSize then Exit;
+    sLowerCase:= PUnicodeChar(Pointer(sSearchText + #0));
+    if bSwapEndian then Utf16SwapEndian(sLowerCase);
+    sLowerCase:= UnicodeLowerCase(sLowerCase);
+
+    // While text size > buffer size
+    while iSize > BUFFER_SIZE do
+    begin
+      SetString(sTextBuffer, PUnicodeChar(pDataAddr + iStartPos), BUFFER_SIZE div 2);
+      if bSwapEndian then Utf16SwapEndian(sTextBuffer);
+      sTextBuffer:= UnicodeLowerCase(sTextBuffer);
+      iTextPos:= Pos(sLowerCase, sTextBuffer);
+      if iTextPos > 0 then
+        Exit(pDataAddr + iStartPos + iTextPos * 2 - 2)
+      else begin
+        // Shift text buffer
+        iStartPos:= iStartPos + (BUFFER_SIZE - iLength);
+      end;
+      iSize:= iDataLength - iStartPos;
+    end;
+    // Process remaining buffer
+    if iLength > iSize then Exit;
+    SetString(sTextBuffer, PUnicodeChar(pDataAddr + iStartPos), iSize div 2);
+    if bSwapEndian then Utf16SwapEndian(sTextBuffer);
+    sTextBuffer:= UnicodeLowerCase(sTextBuffer);
+    iTextPos:= Pos(sLowerCase, sTextBuffer);
+    if iTextPos > 0 then Result:= pDataAddr + iStartPos + iTextPos * 2 - 2;
+  end;
+end;
+
+function FindMmap(const sFileName: String; const sFindData: String;
+  bCase: Boolean; Abort: TAbortFunction): Integer;
 
   function PosMem(pAdr:PChar; iLength:Integer):Pointer;
   var
@@ -278,4 +361,25 @@ begin
   end;
 end;
 
+function FindMmapW(const sFileName: String; const sFindData: String; bLittleEndian: Boolean): Integer;
+var
+  fmr : TFileMapRec;
+begin
+  Result := -1;
+
+  if MapFile(sFileName, fmr) then
+  begin
+    try
+      begin
+        if PosMemW(fmr.MappedFile, fmr.FileSize, 0, sFindData, False, bLittleEndian) <> Pointer(-1) then
+          Result := 1
+        else
+          Result := 0;
+      end;
+    finally
+      UnMapFile(fmr);
+    end;
+  end;
+end;
+
 end.

src/ufindthread.pas

@@ -119,8 +119,17 @@ begin
         FTextSearchType := tsAnsi;
         RecodeTable := InitRecodeTable(TextEncoding, CaseSensitive);
       end
-      else if (CaseSensitive = False) and ((TextEncoding = EncodingUTF8) or (TextEncoding = EncodingUTF8BOM)) then
-        FTextSearchType:= tsUtf8
+      else if (CaseSensitive = False) then
+      begin
+        if ((TextEncoding = EncodingUTF8) or (TextEncoding = EncodingUTF8BOM)) then
+          FTextSearchType:= tsUtf8
+        else if (TextEncoding = EncodingUCS2LE) then
+          FTextSearchType:= tsUtf16le
+        else if (TextEncoding = EncodingUCS2BE) then
+          FTextSearchType:= tsUtf16be
+        else
+          FTextSearchType:= tsOther;
+      end
       else begin
         FTextSearchType:= tsOther;
       end;
@@ -282,9 +291,11 @@ begin
   begin
     // Memory mapping should be slightly faster and use less memory
     case FTextSearchType of
-      tsAnsi:  lastPos:= FindMmapBM(sFileName, sData, RecodeTable, @IsAborting);
-      tsUtf8:  lastPos:= FindMmapU(sFileName, sData)
-      else     lastPos:= FindMmap(sFileName, sData, bCase, @IsAborting);
+      tsAnsi:    lastPos:= FindMmapBM(sFileName, sData, RecodeTable, @IsAborting);
+      tsUtf8:    lastPos:= FindMmapU(sFileName, sData);
+      tsUtf16le: lastPos:= FindMmapW(sFileName, sData, True);
+      tsUtf16be: lastPos:= FindMmapW(sFileName, sData, False);
+      else       lastPos:= FindMmap(sFileName, sData, bCase, @IsAborting);
     end;
     case lastPos of
       0 : Exit(False);
@@ -331,7 +342,13 @@ begin
                 begin
                   if PosMemU(@Buffer[0], DataRead + sDataLength - 1, 0, sData, False) <> Pointer(-1) then
                     Exit(True);
-                end
+                end;
+              tsUtf16le,
+              tsUtf16be:
+                begin
+                  if PosMemW(@Buffer[0], DataRead + sDataLength - 1, 0, sData, False, FTextSearchType = tsUtf16le) <> Pointer(-1) then
+                    Exit(True);
+                end;
               else
                 begin
                   if PosMem(@Buffer[0], DataRead + sDataLength - 1, 0, sData, bCase, False) <> Pointer(-1) then